[PATCH v2] ld: testsuite: Mark vers26b3 unsupported on x86_64

[Jan Beulich]

On 19.08.2025 15:04, Rainer Orth wrote:
> Hi Jan,
> 
>> On 19.08.2025 14:26, Rainer Orth wrote:
>>>> On 19.08.2025 13:55, Rainer Orth wrote:
>>>>> Hi H.J.,
>>>>>
>>>>>> if { [istarget x86_64-*-solaris2*] || [istarget x86_64-*-solaris2*] } then {
>>>>>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>>>>>
>>>>>> Typo.   Should we just do
>>>>>
>>>>> Indeed, thanks.  Alternatively, we could just use [istarget x86_64-*-*]
>>>>> instead: x86_64-*-solaris2* (matched via *-*-solaris2*) and
>>>>> x86_64-*-linux* are the only x86_64 targets for which this test is
>>>>> currently run.
>>>>
>>>> What do you derive this from? I don't see why the test wouldn't also be
>>>> active for x86_64-*-elf, for example. But I guess I'm overlooking something.
>>>
>>> indeed: at the top of vers.exp, the driver just returns unless the
>>> target is listed explicitly.  x86_64-*-elf isn't among them, neither
>>> explicitly or as a wildcard.
>>
>> Hmm, I see. Another anomaly as it looks, considering the i?86-* targets that
> 
> that's because on both Solaris/i386 (once enabled) and Linux/i686 all
> vers tests just PASS.  Only x86_64 is the odd man out, on Linux (19
> XFAIL, 1 UNSUPPORTED) worse than on Solaris (1 UNSUPPORTED).
> 
>> are allowed through. In any event, for the original question - I'd be fine
>> either way, but would prefer the more explicit form.
> 
> Agreed: if we would mark x86_64-*-* unsupported, we'd never notice if
> the test would PASS on some new x86_64 target.
> 
> Should I go with the v2 patch with the typo fixed then?

Yes please.

Jan